Whether it works in Chinese Word Segmentation about lm-lstm-crf

[LiyuanLucasLiu]

Thanks:-)

The current code cannot work for Chinese, but you can definitely modified it a little to make it work on Chinese.

Basically, i would recommended you to use our word-level model, treat each Chinese character as a word, and modify the encoding for read & write. Also, pre-trained embeddings are crucial for performance, i would also recommend you to get some character-level embedding for Chinese (there are several papers about this).

Besides, you could also try to represent Chinese Characters by Wubi or Pingying, and treat them as character-level representation.